diff options
| author | Karl Williamson <khw@cpan.org> | 2020-03-24 18:49:07 -0600 |
|---|---|---|
| committer | Karl Williamson <khw@cpan.org> | 2020-03-24 18:57:20 -0600 |
| commit | f12bbb761071acfd5663d673ad891ffbb2ffdbd9 (patch) | |
| tree | b5e87bd2c873cf4ab27cdad08ac3c6cbfe17e530 /lib/charnames.t | |
| parent | afd41ae4e769eb6f4f106977edabdca9ab36b497 (diff) | |
| download | perl-f12bbb761071acfd5663d673ad891ffbb2ffdbd9.tar.gz | |
Debugging GH #17671.
We are seeing an occasional smoke failure in the experimental regular
expression Unicode property wildcard feature for the name property.
This is not reproducible, and neither valgrind nor asan shows anything.
All the failures I have seen involve looking up NULL. Now it could be
something special about that name, or merely that it is the first in the
test to be checked. This commit seeks to find out by skipping that
test. If we stop getting these intermittent failures, it points one
direction; if they continue with 0x01, instead, it points in another.
Diffstat (limited to 'lib/charnames.t')
| -rw-r--r-- | lib/charnames.t | 4 |
1 files changed, 4 insertions, 0 deletions
diff --git a/lib/charnames.t b/lib/charnames.t index eb98c4576d..030047468b 100644 --- a/lib/charnames.t +++ b/lib/charnames.t @@ -157,6 +157,10 @@ sub test_vianame ($$$) { $wildcard_count++; + # XXX temporary to see if the failure we are occasionally seeing is + # confined to this code point. GH #17671 + next if $i == 0; + # Because wildcard name matching is so real-time intensive, do it less # frequently than the others if ($wildcard_count >= 10) { |